Although there is substantial evidence that dietary fat is positively and dietary fiber negatively associated with colon cancer, some "anomalies" in the epidemiology of this disease have been observed. For example, in Hawaii, Japanese have high rates of colon cancer despite a relatively low fat intake, and Hawaiians experience a low risk despite a high fat intake. Both groups have a similarly low fiber intake. It is hypothesized that these anomalies, as well as other inconsistencies in the epidemiologic literature, may be explained by a protective effect of increased colonic fermentation against colon cancer. The present application will test this hypothesis in a correlation study, and pretest two biochemical markers of fermentation (breath hydrogen and breath methane) for use in epidemiologic studies. Subjects for this project will be randomly selected among population controls newly interviewed for two ongoing case-control studies conducted by our group. End alveolar breath samples will be obtained by having the subjects exhale an end expiratory sample into a 20-ml syringe, 3-hours after dinner on the day of the interview and again within 15 minutes of rising the following morning. These two measurements appear sufficient to correctly rank individuals according to their daily breath hydrogen excretion. Participants will mail in the breath samples for hydrogen and methane analysis by gas-chromatography. Breath hydrogen is known to have a high validity, sensitivity and specificity as a marker of fermentation. It also has a high within-person variability which makes it appropriate for studies at the population level only. Although breath methane appears to be less specific for fermentation, it has been correlated with colon cancer risk among certain populations. The present study will investigate among 350 men and women the correlation between the population means for these markers and colon cancer incidence rates among ten ethnic and sex groups living in Hawaii, as a first step in testing the fermentation hypothesis. If the results are supportive of an association, we plan to seek separate funding to investigate whether these markers can be used in studies at the individual level, by using repeated measurements.